Governed RAG AI
Orchestrates secure Retrieval-Augmented Generation (RAG) with role-based access control, ensuring sensitive data protection and compliance for enterprise knowledge bases.
About
Governed RAG AI is a robust solution addressing the risks of sensitive data exposure in traditional RAG systems. It utilizes Mastra AI orchestration to provide secure Retrieval-Augmented Generation, implementing comprehensive role-based access control (RBAC). The platform boasts hierarchical RBAC with role inheritance, document classification, a multi-agent architecture for secure retrieval, reranking, answering, and verification, and audit-ready logging. Designed for enterprise knowledge bases in sectors like HR, finance, and engineering, it supports multiple LLMs (Gemini, OpenAI, Openrouter) and offers advanced features such as multi-tenant support and step-up authentication for elevated access, all powered by a Next.js frontend and TypeScript backend with Qdrant for vector storage.
Key Features
- Hierarchical Role-Based Access Control (RBAC) with role inheritance
- Multi-agent architecture for secure retrieval, reranking, answering, and verification
- Document classification (public, internal, confidential) with tag-based filtering
- Audit-ready system with citations and logs for compliance (e.g., NIST SP 800-53 AU-2)
- Multi-tenant support and step-up authentication for elevated confidential access
- 1 GitHub stars
Use Cases
- Building secure internal AI assistants for enterprises
- Retrieving departmental knowledge such as finance policies or engineering handbooks
- Facilitating compliant document Q&A within organizations